Terrorism occurs in around 75 per cent of civil conflicts. Previous studies have explained variation in the number of terrorist attacks by rebel group, region, and country per month, year and over longer periods of time. They identify group, country and regime characteristics that make terrorism more likely. However, these measures have not been tested for their predictive performance. This study will make two contributions to the literature on terrorism in civil conflicts: it will test the predictive power of existing indicators used in the literature, and it will not only test predictive performance regarding the incidence but also the onset of terrorism. Support vector machines with radial kernels are used in combination with interpretable machine learning techniques to illustrate findings intuitively and to go beyond standard measures of accuracy, precision and recall. Evidence from available replication data suggests that many variables identified as explanatory factors in previous studies have limited predictive power.